UV and Air Cleaner Installation and Verification

Purpose

A powered device in the airstream gets judged by the customer on whether it lights up. It gets judged by this shop on three things they will never ask about: what it cost in static pressure, whether it puts anything into the air that was not there before, and whether the ultraviolet stays inside the cabinet. Miss the first and the system loses airflow it did not have to spare. Miss the third and the lamp quietly cooks the drain tubing and the blower harness it happens to be pointed at, and the callback arrives two or three years later as a crumbled wire nobody connects to the accessory.

UV-C here means germicidal ultraviolet around 254 nm, used either to keep the evaporator coil clean or to treat air moving through the duct. Those are different jobs with different mounting, and a lamp aimed for one does not do the other.

Scope

Covers installation and verification of in-duct UV-C lamps and in-duct electronic or ionizing air cleaners on residential and small light commercial forced-air systems: selection and listing, baseline measurement, mounting and line-of-sight control, interlock wiring, restoration, pressure verification and handover.

Does not cover media filtration or the pressure budget method itself, which is owned by the filtration upgrade assessment SOP and used here rather than re-derived. Does not cover portable room air cleaners, duct cleaning, or anything sold as an ozone generator for occupied space, which this shop does not install.

Roles and responsibilities

Procedure

1. Confirm the device type, the job it is meant to do, and its listing before anything is opened. Acceptance: device recorded as coil irradiation or in-duct air treatment, plus documentation that it is validated as producing no ozone, or certified under the air cleaner regulation of the state it is installed in where one applies, California being the common case. What wrong looks like: a coil-cleaning device mounted where it treats moving air, which is a different dose problem. Stop rule: no listing documentation, no install. Hazard: none at this step, paperwork read at the truck.

2. Measure baseline total external static and airflow before you cut anything. Acceptance: total external static in in w.c. against the blower's nameplate rating, plus delivered cfm, both with the system in its highest airflow mode and all panels on. What wrong looks like: an install with no before reading, which makes the after reading meaningless. Stop rule: a system already at or over nameplate gets no accessory until the return work is done. Hazard: system running, so probes go through the ports and hands stay out of the cabinet.

3. Isolate power, prove it dead, and pick the opening location. Prove the meter live, prove the conductors dead, prove the meter live again per NFPA 70E-2021, 120.5, with work practices at 29 CFR 1910.333(b)(2), then isolate the blower under 29 CFR 1910.147. Acceptance: 0 volts at the equipment terminals, wheel at rest, and a marked opening meeting the manual's position relative to the coil. What wrong looks like: an opening chosen for access convenience rather than for what the lamp will see. Stop rule: no position that satisfies the manual, no install. Hazard: check what is behind the sheet metal before the bit turns, since a hole through a refrigerant line, a flue or a harness is the same mistake three ways.

4. Cut, mount, and then sight the beam path with the lamp de-energized. Look along the lamp axis and identify everything in direct line of sight. Acceptance: nothing in the beam but metal and the coil itself, with flexible drain tubing, wire insulation, plastic pans, filter media and flex duct either shielded with sheet metal or relocated out of view. What wrong looks like: a lamp with a clear shot down the plenum at the blower harness, which is invisible on the day and brittle in a couple of years. Stop rule: anything organic still in the beam after mounting gets shielded before the lamp is ever energized. Hazard, on the cutting rather than the access: opening a lined plenum liberates fiberglass, so cut with the blower off and eyes protected, and where liner is degraded and shedding the control for that route is a respirator under a written program per 29 CFR 1910.134.

5. Wire the lamp through its interlock and label every access point that can expose it. Acceptance: interlock fitted per the manual so the lamp cannot energize with the access panel off, and a warning label at each panel opening into the irradiated space. What wrong looks like: a lamp wired straight to the line side with no interlock, so the next tech opens a door into a live source. Stop rule: no interlock provision means it is mounted where no routine service access looks into it, or it does not go in. Hazard: never energize a lamp outside its cabinet to check it lights, because UV-C burns cornea and skin at exposures short enough that there is no warning sensation until hours later, and the 254 nm limit in the current ACGIH TLV booklet is small enough that a few unshielded minutes at close range exceeds a day's allowance.

6. Restore power, then prove the interlock actually kills the lamp. Panels on first, then close the disconnect from the hinge side rather than square in front. Acceptance: with the cabinet closed the lamp shows lit at its indicator or viewport, opening the access panel extinguishes it, and no visible violet glow escapes at any seam viewed from a normal standing position. Hazard: this is the step that puts both energy and an optical hazard back into a room with people in it, so confirmation is taken at the indicator or the manufacturer's viewport and nobody looks into the chamber directly, and the customer and pets stay out until the panels are on. Stop rule: an interlock that does not extinguish the lamp is corrected before the job closes, and until then the device is left disconnected.

7. Re-measure static and airflow, and spend the reading against the pressure budget. Acceptance: the increase over the step 2 baseline is at or below half the headroom between that baseline and the nameplate rating, and cfm per ton is still in the 350 to 400 band on a cooling system. That half-headroom rule is the filtration SOP's reserve convention, used here so a filter and an accessory do not both spend the same margin. What wrong looks like: an accessory that fits inside nameplate today and puts the system over it the month the filter loads. Stop rule: over budget, the device comes out or the return work gets quoted. Hazard: readings taken with panels on, probes through the ports only.

8. Hand over with the replacement date, and tell the customer what a problem smells like. Acceptance: the customer can state the month the lamp gets replaced and can say that a sharp, chlorine-like or metallic odor at the registers is a reason to call, not something to wait out. What wrong looks like: a lamp treated as a lifetime part, which keeps glowing well after its useful output has fallen off. Stop rule: if the customer cannot repeat the date back, it goes on a sticker at the air handler as well as on the ticket. Hazard: none at this step, a conversation with the equipment closed up.

When the site does not match

No cabinet position gives a clean line of sight: the device does not go in on the argument that shielding is probably fine, since the shield is the control and a beam path cannot be inspected later. Customer supplied the device: the listing documentation still has to exist first, and if it does not, the shop declines in writing rather than fitting it with a caveat. Accessory found already installed by others: run steps 2, 4, 6 and 7 as an inspection and report a missing interlock or an irradiated harness as a defect rather than repairing it silently.

The record this produces

The install sheet carries: device type, model and serial; the listing or ozone validation document referenced by name; baseline total external static, nameplate rating and baseline cfm; the line-of-sight result with anything shielded or relocated listed by item; the interlock test result, meaning what happened to the lamp when the panel came off; post-install static, the computed increase and the permitted increase; post-install cfm and cfm per ton; and the lamp replacement date.

Two people read it. The next tech, who needs to know before opening a panel that there is a UV source behind it and whether the interlock was proved. And the service manager, if a customer reports an odor or a failed plenum component, because a line-of-sight record naming what was shielded is the difference between a warranty conversation and a guess.

Worked pass: three-ton air handler, coil irradiation lamp added

Customer asked for a coil lamp after a mold complaint the IAQ visit traced to a wet crawlspace, so it is being fitted alongside the real fix, not instead of it.

Step 1. Device recorded as coil irradiation. Manufacturer's zero-ozone validation document on file. Pass.

Step 2. Baseline total external static 0.44 in w.c. against a 0.50 nameplate. Airflow 1,140 cfm; 1,140 divided by 3 tons is 380 cfm per ton. Both recorded.

Step 3. Terminals read 0.0 volts, meter proved live before and after, blower isolated and at rest. Opening marked per the manual at the coil's downstream face.

Step 4. Mounted, then sighted along the lamp axis with the lamp de-energized. In direct line of sight: the flexible condensate tubing where it crosses the pan, and about 8 in of the blower harness. Step 4 fails and takes its stop rule. A sheet metal deflector was fabricated and fitted to block both, then the beam path was re-sighted. Second sighting shows metal and coil only. Pass on retest, with both shielded items listed on the sheet.

Step 5. Interlock fitted to the blower access panel per the manual. Warning labels on the blower panel and the coil access panel.

Step 6. Panels on, disconnect closed from the hinge side. Lamp shows lit at the viewport. Blower access panel opened: lamp extinguishes immediately. No glow visible at any seam from a standing position. Pass.

Step 7. Post-install total external static 0.46 in w.c. The increase is 0.46 minus 0.44, or 0.02 in w.c. Headroom at baseline was 0.50 minus 0.44, or 0.06, so the permitted increase is half of that, 0.03. 0.02 is under 0.03. Airflow re-measured 1,125 cfm; 1,125 divided by 3 is 375 cfm per ton, inside the band. Pass.

Step 8. Replacement date stated by the customer and stickered at the air handler. Odor instruction given and repeated back.

References

• Manufacturer's installation manual for the specific device, for mounting position relative to the coil, interlock provision, rated pressure drop and lamp life
• The current ACGIH TLV booklet for the 254 nm ultraviolet exposure limit, since the numbers are revised on the booklet's own cycle, and EPA guidance on cleaning up a broken mercury-containing lamp, which is the procedure if one breaks: people and pets out, room aired, no vacuum
• The air cleaner certification requirement of the state where the device is installed, California's being the common case, plus any zero-ozone emissions validation the manufacturer holds
• 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5 for live-dead-live; 29 CFR 1910.147 for the blower; 29 CFR 1910.134 for respiratory protection under a written program
• See related: the filtration upgrade assessment SOP, which owns the pressure budget method used at step 7